bson 4.1.1 → 5.2.0

data/spec/spec_tests/data/corpus/top.json

@@ -0,0 +1,262 @@
+{
+    "description": "Top-level document validity",
+    "bson_type": "0x00",
+    "valid": [
+        {
+            "description": "Dollar-prefixed key in top-level document",
+            "canonical_bson": "0F00000010246B6579002A00000000",
+            "canonical_extjson": "{\"$key\": {\"$numberInt\": \"42\"}}"
+        },
+        {
+            "description": "Dollar as key in top-level document",
+            "canonical_bson": "0E00000002240002000000610000",
+            "canonical_extjson": "{\"$\": \"a\"}"
+        },
+        {
+            "description": "Dotted key in top-level document",
+            "canonical_bson": "1000000002612E620002000000630000",
+            "canonical_extjson": "{\"a.b\": \"c\"}"
+        },
+        {
+            "description": "Dot as key in top-level document",
+            "canonical_bson": "0E000000022E0002000000610000",
+            "canonical_extjson": "{\".\": \"a\"}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "An object size that's too small to even include the object size, but is a well-formed, empty object",
+            "bson": "0100000000"
+        },
+        {
+            "description": "An object size that's only enough for the object size, but is a well-formed, empty object",
+            "bson": "0400000000"
+        },
+        {
+            "description": "One object, with length shorter than size (missing EOO)",
+            "bson": "05000000"
+        },
+        {
+            "description": "One object, sized correctly, with a spot for an EOO, but the EOO is 0x01",
+            "bson": "0500000001"
+        },
+        {
+            "description": "One object, sized correctly, with a spot for an EOO, but the EOO is 0xff",
+            "bson": "05000000FF"
+        },
+        {
+            "description": "One object, sized correctly, with a spot for an EOO, but the EOO is 0x70",
+            "bson": "0500000070"
+        },
+        {
+            "description": "Byte count is zero (with non-zero input length)",
+            "bson": "00000000000000000000"
+        },
+        {
+            "description": "Stated length exceeds byte count, with truncated document",
+            "bson": "1200000002666F6F0004000000626172"
+        },
+        {
+            "description": "Stated length exceeds byte count, with valid envelope",
+            "bson": "1300000002666F6F00040000006261720000"
+        },
+        {
+            "description": "Stated length less than byte count, with valid envelope",
+            "bson": "1100000002666F6F00040000006261720000"
+        },
+        {
+            "description": "Invalid BSON type low range",
+            "bson": "07000000000000"
+        },
+        {
+            "description": "Invalid BSON type high range",
+            "bson": "07000000800000"
+        },
+        {
+            "description": "Document truncated mid-key",
+            "bson": "1200000002666F"
+        },
+        {
+            "description": "Null byte in document key",
+            "bson": "0D000000107800000100000000"
+        }
+    ],
+    "parseErrors": [
+        {
+            "description" : "Bad $regularExpression (extra field)",
+            "string" : "{\"a\" : {\"$regularExpression\": {\"pattern\": \"abc\", \"options\": \"\", \"unrelated\": true}}}"
+        },
+        {
+            "description" : "Bad $regularExpression (missing options field)",
+            "string" : "{\"a\" : {\"$regularExpression\": {\"pattern\": \"abc\"}}}"
+        },
+        {
+            "description": "Bad $regularExpression (pattern is number, not string)",
+            "string": "{\"x\" : {\"$regularExpression\" : { \"pattern\": 42, \"options\" : \"\"}}}"
+        },
+        {
+            "description": "Bad $regularExpression (options are number, not string)",
+            "string": "{\"x\" : {\"$regularExpression\" : { \"pattern\": \"a\", \"options\" : 0}}}"
+        },
+        {
+            "description" : "Bad $regularExpression (missing pattern field)",
+            "string" : "{\"a\" : {\"$regularExpression\": {\"options\":\"ix\"}}}"
+        },
+        {
+            "description": "Bad $oid (number, not string)",
+            "string": "{\"a\" : {\"$oid\" : 42}}"
+        },
+        {
+            "description": "Bad $oid (extra field)",
+            "string": "{\"a\" : {\"$oid\" : \"56e1fc72e0c917e9c4714161\", \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $numberInt (number, not string)",
+            "string": "{\"a\" : {\"$numberInt\" : 42}}"
+        },
+        {
+            "description": "Bad $numberInt (extra field)",
+            "string": "{\"a\" : {\"$numberInt\" : \"42\", \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $numberLong (number, not string)",
+            "string": "{\"a\" : {\"$numberLong\" : 42}}"
+        },
+        {
+            "description": "Bad $numberLong (extra field)",
+            "string": "{\"a\" : {\"$numberLong\" : \"42\", \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $numberDouble (number, not string)",
+            "string": "{\"a\" : {\"$numberDouble\" : 42}}"
+        },
+        {
+            "description": "Bad $numberDouble (extra field)",
+            "string": "{\"a\" : {\"$numberDouble\" : \".1\", \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $numberDecimal (number, not string)",
+            "string": "{\"a\" : {\"$numberDecimal\" : 42}}"
+        },
+        {
+            "description": "Bad $numberDecimal (extra field)",
+            "string": "{\"a\" : {\"$numberDecimal\" : \".1\", \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $binary (binary is number, not string)",
+            "string": "{\"x\" : {\"$binary\" : {\"base64\" : 0, \"subType\" : \"00\"}}}"
+        },
+        {
+            "description": "Bad $binary (type is number, not string)",
+            "string": "{\"x\" : {\"$binary\" : {\"base64\" : \"\", \"subType\" : 0}}}"
+        },
+        {
+            "description": "Bad $binary (missing $type)",
+            "string": "{\"x\" : {\"$binary\" : {\"base64\" : \"//8=\"}}}"
+        },
+        {
+            "description": "Bad $binary (missing $binary)",
+            "string": "{\"x\" : {\"$binary\" : {\"subType\" : \"00\"}}}"
+        },
+        {
+            "description": "Bad $binary (extra field)",
+            "string": "{\"x\" : {\"$binary\" : {\"base64\" : \"//8=\", \"subType\" : 0, \"unrelated\": true}}}"
+        },
+        {
+            "description": "Bad $code (type is number, not string)",
+            "string": "{\"a\" : {\"$code\" : 42}}"
+        },
+        {
+            "description": "Bad $code (type is number, not string) when $scope is also present",
+            "string": "{\"a\" : {\"$code\" : 42, \"$scope\" : {}}}"
+        },
+        {
+            "description": "Bad $code (extra field)",
+            "string": "{\"a\" : {\"$code\" : \"\", \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $code with $scope (scope is number, not doc)",
+            "string": "{\"x\" : {\"$code\" : \"\", \"$scope\" : 42}}"
+        },
+        {
+            "description": "Bad $timestamp (type is number, not doc)",
+            "string": "{\"a\" : {\"$timestamp\" : 42} }"
+        },
+        {
+            "description": "Bad $timestamp ('t' type is string, not number)",
+            "string": "{\"a\" : {\"$timestamp\" : {\"t\" : \"123456789\", \"i\" : 42} } }"
+        },
+        {
+            "description": "Bad $timestamp ('i' type is string, not number)",
+            "string": "{\"a\" : {\"$timestamp\" : {\"t\" : 123456789, \"i\" : \"42\"} } }"
+        },
+        {
+            "description": "Bad $timestamp (extra field at same level as $timestamp)",
+            "string": "{\"a\" : {\"$timestamp\" : {\"t\" : \"123456789\", \"i\" : \"42\"}, \"unrelated\": true } }"
+        },
+        {
+            "description": "Bad $timestamp (extra field at same level as t and i)",
+            "string": "{\"a\" : {\"$timestamp\" : {\"t\" : \"123456789\", \"i\" : \"42\", \"unrelated\": true} } }"
+        },
+        {
+            "description": "Bad $timestamp (missing t)",
+            "string": "{\"a\" : {\"$timestamp\" : {\"i\" : \"42\"} } }"
+        },
+        {
+            "description": "Bad $timestamp (missing i)",
+            "string": "{\"a\" : {\"$timestamp\" : {\"t\" : \"123456789\"} } }"
+        },
+        {
+            "description": "Bad $date (number, not string or hash)",
+            "string": "{\"a\" : {\"$date\" : 42}}"
+        },
+        {
+            "description": "Bad $date (extra field)",
+            "string": "{\"a\" : {\"$date\" : {\"$numberLong\" : \"1356351330501\"}, \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $minKey (boolean, not integer)",
+            "string": "{\"a\" : {\"$minKey\" : true}}"
+        },
+        {
+            "description": "Bad $minKey (wrong integer)",
+            "string": "{\"a\" : {\"$minKey\" : 0}}"
+        },
+        {
+            "description": "Bad $minKey (extra field)",
+            "string": "{\"a\" : {\"$minKey\" : 1, \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad $maxKey (boolean, not integer)",
+            "string": "{\"a\" : {\"$maxKey\" : true}}"
+        },
+        {
+            "description": "Bad $maxKey (wrong integer)",
+            "string": "{\"a\" : {\"$maxKey\" : 0}}"
+        },
+        {
+            "description": "Bad $maxKey (extra field)",
+            "string": "{\"a\" : {\"$maxKey\" : 1, \"unrelated\": true}}"
+        },
+        {
+            "description": "Bad DBpointer (extra field)",
+            "string": "{\"a\": {\"$dbPointer\": {\"a\": {\"$numberInt\": \"1\"}, \"$id\": {\"$oid\": \"56e1fc72e0c917e9c4714161\"}, \"c\": {\"$numberInt\": \"2\"}, \"$ref\": \"b\"}}}"
+        },
+        {
+            "description" : "Null byte in document key",
+            "string" : "{\"a\\u0000\": 1 }"
+        },
+        {
+            "description" : "Null byte in sub-document key",
+            "string" : "{\"a\" : {\"b\\u0000\": 1 }}"
+        },
+        {
+            "description": "Null byte in $regularExpression pattern",
+            "string": "{\"a\" : {\"$regularExpression\" : { \"pattern\": \"b\\u0000\", \"options\" : \"i\"}}}"
+        },
+        {
+            "description": "Null byte in $regularExpression options",
+            "string": "{\"a\" : {\"$regularExpression\" : { \"pattern\": \"b\", \"options\" : \"i\\u0000\"}}}"
+        }
+    ]
+}

data/spec/spec_tests/data/corpus/undefined.json

@@ -0,0 +1,15 @@
+{
+    "description": "Undefined type (deprecated)",
+    "bson_type": "0x06",
+    "deprecated": true,
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Undefined",
+            "canonical_bson": "0800000006610000",
+            "canonical_extjson": "{\"a\" : {\"$undefined\" : true}}",
+            "converted_bson": "080000000A610000",
+            "converted_extjson": "{\"a\" : null}"
+        }
+    ]
+}

data/spec/spec_tests/data/corpus_legacy/array.json

@@ -0,0 +1,49 @@
+{
+    "description": "Array",
+    "bson_type": "0x04",
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Empty",
+            "bson": "0D000000046100050000000000",
+            "extjson": "{\"a\" : []}"
+        },
+        {
+            "description": "Single Element Array",
+            "bson": "140000000461000C0000001030000A0000000000",
+            "extjson": "{\"a\" : [10]}"
+        },
+        {
+            "description": "Single Element Array with index set incorrectly to empty string",
+            "bson": "130000000461000B00000010000A0000000000",
+            "canonical_bson": "140000000461000C0000001030000A0000000000",
+            "extjson": "{\"a\" : [10]}"
+        },
+        {
+            "description": "Single Element Array with index set incorrectly to ab",
+            "bson": "150000000461000D000000106162000A0000000000",
+            "canonical_bson": "140000000461000C0000001030000A0000000000",
+            "extjson": "{\"a\" : [10]}"
+        },
+        {
+            "description": "Multi Element Array with duplicate indexes",
+            "bson": "1b000000046100130000001030000a000000103000140000000000",
+            "canonical_bson": "1b000000046100130000001030000a000000103100140000000000",
+            "extjson": "{\"a\" : [10, 20]}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Array length too long: eats outer terminator",
+            "bson": "140000000461000D0000001030000A0000000000"
+        },
+        {
+            "description": "Array length too short: leaks terminator",
+            "bson": "140000000461000B0000001030000A0000000000"
+        },
+        {
+            "description": "Invalid Array: bad string length in field",
+            "bson": "1A00000004666F6F00100000000230000500000062617A000000"
+        }
+    ]
+}

data/spec/spec_tests/data/corpus_legacy/binary.json

@@ -0,0 +1,69 @@
+{
+    "description": "Binary type",
+    "bson_type": "0x05",
+    "test_key": "x",
+    "valid": [
+        {
+            "description": "subtype 0x00 (Zero-length)",
+            "bson": "0D000000057800000000000000",
+            "extjson": "{\"x\" : {\"$binary\" : \"\", \"$type\" : \"00\"}}"
+        },
+        {
+            "description": "subtype 0x00",
+            "bson": "0F0000000578000200000000FFFF00",
+            "extjson": "{\"x\" : {\"$binary\" : \"//8=\", \"$type\" : \"00\"}}"
+        },
+        {
+            "description": "subtype 0x01",
+            "bson": "0F0000000578000200000001FFFF00",
+            "extjson": "{\"x\" : {\"$binary\" : \"//8=\", \"$type\" : \"01\"}}"
+        },
+        {
+            "description": "subtype 0x02",
+            "bson": "13000000057800060000000202000000ffff00",
+            "extjson": "{\"x\" : {\"$binary\" : \"//8=\", \"$type\" : \"02\"}}"
+        },
+        {
+            "description": "subtype 0x03",
+            "bson": "1D000000057800100000000373FFD26444B34C6990E8E7D1DFC035D400",
+            "extjson": "{\"x\" : {\"$binary\" : \"c//SZESzTGmQ6OfR38A11A==\", \"$type\" : \"03\"}}"
+        },
+        {
+            "description": "subtype 0x04",
+            "bson": "1D000000057800100000000473FFD26444B34C6990E8E7D1DFC035D400",
+            "extjson": "{\"x\" : {\"$binary\" : \"c//SZESzTGmQ6OfR38A11A==\", \"$type\" : \"04\"}}"
+        },
+        {
+            "description": "subtype 0x05",
+            "bson": "1D000000057800100000000573FFD26444B34C6990E8E7D1DFC035D400",
+            "extjson": "{\"x\" : {\"$binary\" : \"c//SZESzTGmQ6OfR38A11A==\", \"$type\" : \"05\"}}"
+        },
+        {
+            "description": "subtype 0x80",
+            "bson": "0F0000000578000200000080FFFF00",
+            "extjson": "{\"x\" : {\"$binary\" : \"//8=\", \"$type\" : \"80\"}}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Length longer than document",
+            "bson": "1D000000057800FF0000000573FFD26444B34C6990E8E7D1DFC035D400"
+        },
+        {
+            "description": "Negative length",
+            "bson": "0D000000057800FFFFFFFF0000"
+        },
+        {
+            "description": "subtype 0x02 length too long ",
+            "bson": "13000000057800060000000203000000FFFF00"
+        },
+        {
+            "description": "subtype 0x02 length too short",
+            "bson": "13000000057800060000000201000000FFFF00"
+        },
+        {
+            "description": "subtype 0x02 length negative one",
+            "bson": "130000000578000600000002FFFFFFFFFFFF00"
+        }
+    ]
+}

data/spec/spec_tests/data/corpus_legacy/boolean.json

@@ -0,0 +1,27 @@
+{
+    "description": "Boolean",
+    "bson_type": "0x08",
+    "test_key": "b",
+    "valid": [
+        {
+            "description": "True",
+            "bson": "090000000862000100",
+            "extjson": "{\"b\" : true}"
+        },
+        {
+            "description": "False",
+            "bson": "090000000862000000",
+            "extjson": "{\"b\" : false}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Invalid boolean value of 2",
+            "bson": "090000000862000200"
+        },
+        {
+            "description": "Invalid boolean value of -1",
+            "bson": "09000000086200FF00"
+        }
+    ]
+}

data/spec/spec_tests/data/corpus_legacy/code.json

@@ -0,0 +1,67 @@
+{
+    "description": "Code",
+    "bson_type": "0x0D",
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Empty string",
+            "bson": "0D0000000D6100010000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\"}}"
+        },
+        {
+            "description": "Single character",
+            "bson": "0E0000000D610002000000620000",
+            "extjson": "{\"a\" : {\"$code\" : \"b\"}}"
+        },
+        {
+            "description": "Multi-character",
+            "bson": "190000000D61000D0000006162616261626162616261620000",
+            "extjson": "{\"a\" : {\"$code\" : \"abababababab\"}}"
+        },
+        {
+            "description": "two-byte UTF-8 (\u00e9)",
+            "bson": "190000000261000D000000C3A9C3A9C3A9C3A9C3A9C3A90000",
+            "extjson": "{\"a\" : \"\\u00e9\\u00e9\\u00e9\\u00e9\\u00e9\\u00e9\"}"
+        },
+        {
+            "description": "three-byte UTF-8 (\u2606)",
+            "bson": "190000000261000D000000E29886E29886E29886E298860000",
+            "extjson": "{\"a\" : \"\\u2606\\u2606\\u2606\\u2606\"}"
+        },
+        {
+            "description": "Embedded nulls",
+            "bson": "190000000261000D0000006162006261620062616261620000",
+            "extjson": "{\"a\" : \"ab\\u0000bab\\u0000babab\"}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "bad code string length: 0 (but no 0x00 either)",
+            "bson": "0C0000000261000000000000"
+        },
+        {
+            "description": "bad code string length: -1",
+            "bson": "0C000000026100FFFFFFFF00"
+        },
+        {
+            "description": "bad code string length: eats terminator",
+            "bson": "10000000026100050000006200620000"
+        },
+        {
+            "description": "bad code string length: longer than rest of document",
+            "bson": "120000000200FFFFFF00666F6F6261720000"
+        },
+        {
+            "description": "code string is not null-terminated",
+            "bson": "1000000002610004000000616263FF00"
+        },
+        {
+            "description": "empty code string, but extra null",
+            "bson": "0E00000002610001000000000000"
+        },
+        {
+            "description": "invalid UTF-8",
+            "bson": "0E00000002610002000000E90000"
+        }
+    ]
+}

data/spec/spec_tests/data/corpus_legacy/code_w_scope.json

@@ -0,0 +1,78 @@
+{
+    "description": "CodeWithScope",
+    "bson_type": "0x0F",
+    "test_key": "a",
+    "valid": [
+        {
+            "description": "Empty code string, empty scope",
+            "bson": "160000000F61000E0000000100000000050000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\", \"$scope\" : {}}}"
+        },
+        {
+            "description": "Non-empty code string, empty scope",
+            "bson": "1A0000000F610012000000050000006162636400050000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"abcd\", \"$scope\" : {}}}"
+        },
+        {
+            "description": "Empty code string, non-empty scope",
+            "bson": "1D0000000F61001500000001000000000C000000107800010000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\", \"$scope\" : {\"x\" : 1}}}"
+        },
+        {
+            "description": "Non-empty code string and non-empty scope",
+            "bson": "210000000F6100190000000500000061626364000C000000107800010000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"abcd\", \"$scope\" : {\"x\" : 1}}}"
+        },
+        {
+            "description": "Unicode and embedded null in code string, empty scope",
+            "bson": "1A0000000F61001200000005000000C3A9006400050000000000",
+            "extjson": "{\"a\" : {\"$code\" : \"\\u00e9\\u0000d\", \"$scope\" : {}}}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "field length zero",
+            "bson": "280000000F6100000000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length negative",
+            "bson": "280000000F6100FFFFFFFF0500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length too short (less than minimum size)",
+            "bson": "160000000F61000D0000000100000000050000000000"
+        },
+        {
+            "description": "field length too short (truncates scope)",
+            "bson": "280000000F61001F0000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length too long (clips outer doc)",
+            "bson": "280000000F6100210000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "field length too long (longer than outer doc)",
+            "bson": "280000000F6100FF0000000500000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: length too short",
+            "bson": "280000000F6100200000000400000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: length too long (clips scope)",
+            "bson": "280000000F6100200000000600000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: negative length",
+            "bson": "280000000F610020000000FFFFFFFF61626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad code string: length longer than field",
+            "bson": "280000000F610020000000FF00000061626364001300000010780001000000107900010000000000"
+        },
+        {
+            "description": "bad scope doc (field has bad string length)",
+            "bson": "1C0000000F001500000001000000000C000000020000000000000000"
+        }
+    ]
+}

data/spec/spec_tests/data/corpus_legacy/document.json

@@ -0,0 +1,36 @@
+{
+    "description": "Document",
+    "bson_type": "0x03",
+    "test_key": "x",
+    "valid": [
+        {
+            "description": "Empty subdoc",
+            "bson": "0D000000037800050000000000",
+            "extjson": "{\"x\" : {}}"
+        },
+        {
+            "description": "Empty-string key subdoc",
+            "bson": "150000000378000D00000002000200000062000000",
+            "extjson": "{\"x\" : {\"\" : \"b\"}}"
+        },
+        {
+            "description": "Single-character key subdoc",
+            "bson": "160000000378000E0000000261000200000062000000",
+            "extjson": "{\"x\" : {\"a\" : \"b\"}}"
+        }
+    ],
+    "decodeErrors": [
+        {
+            "description": "Subdocument length too long: eats outer terminator",
+            "bson": "1800000003666F6F000F0000001062617200FFFFFF7F0000"
+        },
+        {
+            "description": "Subdocument length too short: leaks terminator",
+            "bson": "1500000003666F6F000A0000000862617200010000"
+        },
+        {
+            "description": "Invalid subdocument: bad string length in field",
+            "bson": "1C00000003666F6F001200000002626172000500000062617A000000"
+        }
+    ]
+}